What's in your Monday briefing
The Monday briefing has seven sections, in the same order every week: Headline, Revenue pulse, Usage signals, People to talk to, Feedback themes, Watchlist, and What changed. About 400 words. Three minutes to scan.
The shape never changes so your eye learns where to find each thing. Every design decision in Herald is judged against this page: does it make the Monday briefing more valuable?
1. Headline
Section titled “1. Headline”One sentence summarizing the week. Names a specific person or a specific dollar figure when the week earned one.
“Helpkit — 2 churn risks, 1 pricing signal, and why Bob at Acme should hear from you.”
2. Revenue pulse
Section titled “2. Revenue pulse”New MRR, churn, expansion, forecast vs. last 4 weeks. One paragraph. Drawn from Stripe events by the Revenue sub-agent, which queries your DO’s Stripe tables and daily MRR rollup.
3. Usage signals
Section titled “3. Usage signals”Feature adoption deltas, drop-offs, cohorts lighting up. Drawn from your SDK events (in Analytics Engine) and the daily rollups Herald maintains in your Durable Object.
4. People to talk to
Section titled “4. People to talk to”Up to 5 named users — new power users, at-risk accounts, high-signal feedback authors. Each entry is a sentence: who they are, what changed, and one action.
Bob Sterling, Acme Corp — Converted Jan 14, expanded to Team plan Mar 2. This week: 4× usage drop, opened 2 support tickets about Zendesk import. Worth a personal email.
User names link to a side drawer with their full profile.
5. Feedback clusters
Section titled “5. Feedback clusters”Top 3 themes with representative quotes, deduped against the last 4 weeks. Drawn from Intercom, email forwards, sales-call transcripts, reviews — everything that carries text. The Feedback sub-agent clusters via Vectorize embeddings so you don’t see “users want dark mode” for the fifth week in a row.
6. Anomalies
Section titled “6. Anomalies”Statistically significant changes — Welch’s t-test vs. a 28-day baseline, p < 0.05. Stuff Herald noticed that you probably didn’t.
7. Suggested actions
Section titled “7. Suggested actions”Three to five concrete asks. Each one links back to the evidence that produced it.
Who wrote it
Section titled “Who wrote it”Six systems run in parallel every Monday morning: Revenue reads Stripe, Usage reads your events, Feedback reads forwarded emails and reviews, People flags accounts to talk to, Anomaly finds what changed, and Watchlist checks your standing alerts. A final pass (the Composer) stitches those findings into one narrative, matched to the voice of your last four briefings.
The whole thing takes ten to twelve seconds.
Where it’s delivered
Section titled “Where it’s delivered”- Email — primary surface. Reply to follow up; replies route back to chat.
- Web app — the same briefing, rendered with inline charts and drawer navigation.
- Slack — if you’ve connected it (Starter tier and up).
Change cadence, timezone, or delivery channel under Settings → Briefing.